Borderline personality disorder (BPD) is a well-characterized psychiatric disorder with an estimated prevalence of 1-2 percent of the U.S. population, and high levels of impairment and mortality. BPD accounts for approximately 10 percent of all psychiatric outpatients and 20 percent of acute inpatient hospitalizations. Core symptoms of BPD include mood instability, anger, depression, psychoticism/paranoid ideation, aggression and anxiety. In an open label study, we treated 11 BPD patients with the mood stabilizer divalproex sodium (DS). Of the 8 patients who completed treatment, 50 percent showed significant improvement. There was a significant decrease in the SCL-90 total score from 99.1 (q68.4) to 56.6 (q59.5)(t=2.8, p=.03). Global subjective irritability showed significant decreases as well from 3.4(q0.55) to 1.7(q1.4)(t=3.0, p=.02). Improvement was also seen on clinical measures of irritability, anger, anxiety and rejection sensitivity. In a pilot double-blind placebo controlled trial of DS in 16 BPD patients, endpoint CGI improvement scores were 2.0 (much improved) for DS and 5.0 (minimally worse) for placebo. There was significant improvement on global functioning by GAS score (F=5.4, df=1,12, p=.04), aggression by OAS-M (F=19.1, df=1,18, p=.005) and improvement on depression by BDI from 15.5 to 6.9 on DS vs 14.7 to 12 on placebo. This and other preliminary studies suggest potential efficacy of mood stabilizers such as DS in BPD, which would complement existing nonpharmacologic treatment modalities. To further assess the efficacy of DS in BPD in the proposed small grant, 25 outpatients will be randomized to 10 week double-blind treatment with DS vs placebo, preceded by a two-week placebo run in to establish baseline levels on assessment measures. Clinical assessment of symptoms and behaviors by a treating psychiatrist will occur on a weekly basis for the first two weeks following initiation of treatment, then on an every-other week basis for the following 8 weeks. An independent evaluator, blind to dosage and side effects, will conduct assessments every 2 weeks for global severity, psychoticism/paranoid ideation, anger, anxiety and depression. Plasma valproate levels will be monitored, and DS raised to a target value of 70-120 mug/ml. This small grant (R03) will provide systematic information on the efficacy of DS in the treatment of outpatients with BPD, as well as provide information on feasibility, power, response rates, and reliability of outcome measures which will form the basis of a larger RO1 proposal.